Re: DTD Confusion

To put it in short words: I totally agree with Juergen!!!

We had a lot of trouble during implementing our WebDAV client which were
caused by the fact that client and server implementers had different
understandings of the protocol. The majority of that trouble could have been
avoided with a crystal clear DTD (e. g. without or at least with a reduced
numbers of ANYs) because the likelihood of misunderstandings of the protocol
can be tremendously reduced if such a DTD exists.

Best, Hartmut

Juergen Reuter wrote:

> > I also agree that DTDs can and should be avoided. Let's not forget that
> > DTDs originated in an era when most SGML documents were created by
> > humans. WebDAV clients and servers do not need to deal with DTDs.
>
> WebDAV/DeltaV clients and servers are also created by humans.  And the
> WebDAV and DeltaV protocol are created by humans.  I already found a
> couple of bugs in the WebDAV protocol specification just by trying to
> validate the examples that are given in the protocol (just have a look at
> the open issues list for WebDAV!).  And there are frequent messages on
> (at least) the WebDAV versioning list about strange message content
> produced by various servers and clients.  I am pretty sure that, if we
> were using valid XML, we would have much less trouble.
>
> Please note that (as far as I know) no one is trying to force
> implementers to validate WebDAV/DeltaV XML content.  What is requested
> here, is to give implementers the possibility to validate the content in
> a way that conforms with the XML specification, if they choose to do some
> validation.
>
> > Any "invalid" XML within requests or responses indicate (at best) a
> > lack of conformance to specifications or (at worst) a bug
>
> A bug in the specification itself is even worse, since it may lead to
> different opinions about which behaviour is conformant.  And, recognizing
> all those messages about problems with WebDAV clients and servers, bugs
> in WebDAV client/server applications seem to be anything but seldom; so
> it is very likely that you really want some kind of validation.
>
> Greetings,
>            Juergen

Received on Thursday, 8 February 2001 10:15:40 UTC